Interaction with Cfd1 Increases the Kinetic Lability of FeS on the Nbp35 Scaffold*

Background: A Cfd1 and Nbp35 heterocomplex serves as scaffold for cytosolic iron-sulfur cluster assembly. Results: Deficiency in Cfd1-Nbp35 interaction impaired iron turnover on Nbp35. Conclusion: Cfd1 promotes binding and transfer of labile iron-sulfur cluster on the Nbp35 scaffold. Significance: This is the first insight into the unique roles of these P-loop ATPases in cytosolic iron-sulfur cluster assembly. P-loop NTPases of the ApbC/Nbp35 family are involved in FeS protein maturation in nearly all organisms and are proposed to function as scaffolds for initial FeS cluster assembly. In yeast and animals, Cfd1 and Nbp35 are homologous P-loop NTPases that form a heterotetrameric complex essential for FeS protein maturation through the cytosolic FeS cluster assembly (CIA) pathway. Cfd1 is conserved in animals, fungi, and several archaeal species, but in many organisms, only Nbp35 is present, raising the question of the unique roles played by Cfd1 and Nbp35. To begin to investigate this issue, we examined Cfd1 and Nbp35 function in budding yeast. About half of each protein was detected in a heterocomplex in logarithmically growing yeast. Nbp35 readily bound 55Fe when fed to cells, whereas 55Fe binding by free Cfd1 could not be detected. Rapid 55Fe binding to and release from Nbp35 was impaired by Cfd1 deficiency. A Cfd1 mutation that caused a defect in heterocomplex stability supported iron binding to Nbp35 but impaired iron release. Our results suggest a model in which Cfd1-Nbp35 interaction increases the lability of assembled FeS on the Nbp35 scaffold for transfer to target apo-FeS proteins.

Iron-sulfur (FeS) proteins are involved in a wide variety of cellular functions, several essential to life itself (1). The compartmental nature of the eukaryotic cell coupled with the variety and ubiquitous distribution of iron-sulfur (FeS) proteins necessitates multiple systems to ensure efficient FeS protein maturation throughout the cell. All FeS cluster biogenesis in the eukaryotic cell requires the activity of the mitochondrial iron-sulfur cluster (ISC) 5 assembly system (2). Although small amounts of some components of the ISC system have been detected in the cytoplasm of animal cells (3)(4)(5), most cytosolic and nuclear FeS proteins require the action of a cytosolic iron sulfur cluster assembly (CIA) system for their maturation (1, 6 -10). The CIA system is restricted to the cytoplasm and consists of six core proteins, in yeast called Cfd1, Nbp35, Nar1, Cia1, Dre2, and Tah18 (1, 6 -11). CIA is linked to the ISC system through a yet undefined product that is exported out of mitochondria via the ISC export system (2,12).
Cfd1 and Nbp35 are P-loop NTPases that are thought to function as scaffolds for the initial assembly of FeS clusters in the CIA pathway (13,14). These homologous proteins are members of the ApbC/Nbp35 subfamily of P-loop NTPases that is characterized by a conserved C-terminal ENMS sequence followed by a CX 2 C cysteine cluster (15,16). This C-terminal cysteine cluster confers the ability onto these proteins of binding a bridging [4Fe-4S] cluster between monomers and is essential for activity of both proteins (6,14). Nbp35 also has a cysteine cluster at the N terminus, a feature that distinguishes it from Cfd1 and allows the Nbp35 monomer to bind an additional [4Fe-4S] cluster (7,14). Cfd1 and Nbp35 form a heterotetrameric complex that upon cluster reconstitution in vitro binds up to four [4Fe-4S] clusters, two clusters bridging monomers, and one cluster coordinated at the N terminus of each Nbp35 monomer (14). It is currently unknown whether the bridging clusters are between a homodimer or heterodimer within the heterotetrameric arrangement.
The FeS clusters that assembled in vitro on the Cfd1-Nbp35 heterotetramer, or on each protein independently, were readily transferred to target proteins, supporting the view that these CIA factors serve as scaffolds for initial FeS cluster assembly (13). The ability to coordinate FeS cluster and donate cluster to apo target proteins is a conserved feature of members of the ApbC/Nbp35 family. Ind1 in mitochondria of mammals (17), ApbC in bacteria and archaea (18,19), and chloroplast HFC101 (20) and AtNBP35 (21) in plants were each shown to coordinate and transfer reconstituted FeS clusters in vitro. Interestingly, in vitro assembly and transfer of FeS clusters on these P-loop NTPases did not require nucleotide binding or hydrolysis. However, nucleotide binding and hydrolysis are required for iron binding to Cfd1 and Nbp35 in vivo (14).
Members of the ApbC/Nbp35 family are widely distributed, being found in virtually all organisms in the biosphere (17,20,22). Intriguingly, the requirement for two such P-loop NTPases within the same pathway for FeS cluster biogenesis to date has only been demonstrated in animals and fungi (23). Cfd1 is absent in plants and bacteria. The fact that Nbp35 can act alone in a wide range of organisms raises the question of the unique role of Cfd1 and the need for two P-loop NTPases for cytosolic FeS cluster assembly in animals and fungi. Here we investigated this question by examining the role of Cfd1 and Nbp35 in budding yeast. Our results suggest a model for Cfd1 function in which its interaction with Nbp35 alters the character of Nbp35-bound FeS, making it more labile and enhancing transfer to apo target FeS proteins.

EXPERIMENTAL PROCEDURES
Strains, Plasmids, Media, and Growth Conditions-The 0615d strain (MATa, ura3-52, trp1-⌬63, his3-⌬200, aco1-1, ade2, IDP2 up ) was the parental strain used throughout this study and is described elsewhere (24). To generate ⌬cfd1 and ⌬nbp35 strains, the chromosomal copy of each gene was deleted in merodiploid strains using the one-step gene disruption method (25). Briefly, 0615d was transformed with either CFD1 or NBP35 on a CEN/ARS plasmid carrying a URA3 selectable marker (pRS316 (26)). CFD1 or NBP35 on the chromosome was then deleted by targeted gene disruption using TRP1 (26) for CFD1 (deletion from 289 nucleotides upstream to 406 nucleotides downstream of the translation start codon) or a KanMX cassette (27) for NBP35 (deletion of the entire ORF). Gene disruptions were confirmed by PCR amplification of the corresponding chromosomal locus. To construct strains carrying specific CFD1 or NBP35 mutants, the deletion strains were transformed with the indicated mutant gene on a CEN/ARS plasmid (26) followed by counter selection on medium supplemented with 5-fluoroorotic acid (1 mg/ml), selecting for strains that lose the plasmid carrying the wild-type gene (28). CFD1 and NBP35 mutant genes were constructed by site-directed mutagenesis employing a two-step PCR approach (29). Sequencing was performed at the University of Illinois at Chicago Research Resource Center sequencing facility and was compared with published sequences found in the Saccharomyces Genome Database. Yeasts transformations followed the lithium acetate method (30). Transformed yeast cells were grown at 30°C in minimal medium supplemented with 2% dextrose (SD (31)) and lacking nutrients as necessary for selection and maintenance of specific plasmids. Yeast were grown to mid log phase (A 600 ϳ0.8) for all labeling and coimmunoprecipitation experiments.
Immunoprecipitation-Proteins of interest were expressed with a C-terminal epitope tag, and immunoprecipitations were performed using antibodies specific to the epitope tag. Epitopetagged versions of proteins were determined to be functional at a level comparable with untagged proteins by gene complementation analysis. For immunoprecipitation, yeast expressing epitope-tagged proteins were pelleted, washed, and lysed by vigorous agitation with glass beads (0.5 mm) in YBB/EDTA (50 mM Tris-Cl, pH 8, 50 mM KCl, 10% glycerol, and 1 mM EDTA) supplemented with 1 mM phenylmethylsulfonyl fluoride (PMSF). Crude cell lysates were cleared by centrifugation at 16,000 relative centrifugal force for 3 min. Protein concentrations of the cleared cell lysates were determined spectrophotometrically (32) and adjusted to equal protein concentration with YBB/EDTA supplemented with Triton X-100 (0.1% final concentration). The cleared cell extract (1.25 mg of total protein) was then incubated with the indicated monoclonal antibody (Cell Signaling) and protein G agarose beads (Santa Cruz Biotechnology) for 2 h at 4°C. Agarose beads were collected by centrifugation at 1000 relative centrifugal force for 30 s and washed three times with YBB/EDTA ϩ 0.1% Triton X-100. SDS-PAGE sample buffer was added to the washed beads, and half of each reaction was evaluated for precipitated and co-precipitated protein by Western blot. The relative amount of protein precipitated was determined from images of immunostained filters by densitometry using an Alpha Imager (Alpha Innotech) with AlphEaseFC software. Each protein was immunoprecipitated with greater than 90% efficiency as determined from the protein remaining in the post-immunoprecipitation extracts as compared with untreated extracts.
To investigate the amount of Cfd1 or Nbp35 that was engaged in the heterocomplex, ϳ80% of the supernatants from the first (primary) immunoprecipitation was subjected to a second round of immunoprecipitation with antibody specific to epitope-tagged Cfd1 or Nbp35 or was mock precipitated with protein G-agarose beads alone. The resulting immunoprecipitates were analyzed for precipitated protein by Western blot or associated 55 Fe (see below) by liquid scintillation.

55
Fe Labeling and Protein-associated Iron Stability Analysis-Yeast were labeled to steady state with 55 Fe by growing cells for four doublings in SD medium containing 55 FeCl 3 (1 Ci/ml, ϳ1.2 M total iron concentration). To determine the stability of protein-bound 55 Fe, 100 M 2,2Ј-bipyridyl (BIP; Sigma-Aldrich) was added directly to media, and aliquots were removed for measurement of protein-bound 55 Fe at various points after BIP addition. Labeled cells were harvested, washed two times with 1 mM EDTA, washed once with YBB/EDTA, flash-frozen in a dry ice/ethanol bath, and stored at Ϫ80°C until processed. For analysis of protein-bound 55 Fe, cell lysate preparation and immunoprecipitation were performed in an anaerobic glove box using oxygenfree buffers. Immunoprecipitations were performed with 2.5 mg of total extract protein as described above.
For pulse labeling, cells were first resuspended in 0.5ϫ the original volume of SD medium containing 100 M BIP and allowed to grow for 3 h to deplete available cellular iron and stimulate the high affinity iron transport system (33). Cells were collected, washed two times with H 2 O at room temperature, and resuspended in 0.1ϫ the original volume of iron-free SD medium (Q.BIOgene) supplemented with 1 Ci/ml 55 FeCl 3 (ϳ0.1 M iron) and 1 mM ascorbic acid. Aliquots of cells were harvested at designated time points and added to an equal volume of crushed ice to stop cellular iron uptake. Cells were washed and lysed, and protein-bound 55 Fe was evaluated by immunoprecipitation of specific proteins. For pulse-chase experiments, cells were pulse-labeled for 30 min as described above followed by chase with 100 M Fe(NH 4 ) 2 (SO 4 ) 2 , also in the presence of 1 mM ascorbic acid.
To investigate the stability of Nbp35-bound 55 Fe in cfd1 mutant strains, yeast were grown overnight in iron-free medium, at which point cells were collected, washed, and resuspended into 0.1ϫ the original volume of iron-free medium supplemented with 1 Ci/ml 55 FeCl 3 (ϳ1 M iron) and 1 mM ascorbic acid. Cells were allowed to incorporate 55 Fe for 30 min, at which time vehicle or BIP was added as indicated in Fig. 8, and incubation continued a further 30 min. Nbp35 was immunoprecipitated from cleared cell extracts, and 55 Fe was measured by liquid scintillation.

RESULTS
Heterocomplex Formation by Cfd1 and Nbp35-Cfd1 and Nbp35 interact, forming a heterotetrameric complex that binds up to four [4Fe-4S] clusters upon reconstitution in vitro (13,14). It was of interest to determine the fraction of each protein engaged in such a heterocomplex within the yeast cell. To this end, yeast expressing epitope-tagged versions of each protein were grown to mid-log phase in medium supplemented with 55 FeCl 3 and harvested, and the cleared cytoplasmic extracts were subjected to immunoprecipitation of Cfd1 or Nbp35 to Ն90% recovery (Fig. 1A). Each precipitated sample was then analyzed for the presence of both proteins by immunoblot (see "Experimental Procedures"). Comparison of the quantity of each protein recovered by direct immunoprecipitation with the quantity of that protein co-precipitated with the other factor gave the approximate fraction of each protein that was engaged in a heterocomplex. Using this approach, ϳ40% of Cfd1 co-precipitated with Nbp35 and 38% of Nbp35 co-precipitated with Cfd1, whereas the larger portion of each protein was not in a com-plex with the other (Fig. 1A). The extent of co-precipitation did not change after prolonged incubation of cell extracts under the conditions used to perform these precipitations, indicating that this was a stable interaction. This distribution was not altered by starving yeast of iron or by growing yeast in medium with excess iron. 6 Further, co-precipitation of these proteins was not detected after mixing extracts from cells independently expressing epitopetagged Cfd1 or Nbp35, showing that association of these proteins did not occur in the cell-free extracts (not shown).
Iron Distribution between Cfd1, Nbp35, and the Cfd1-Nbp35 Heterocomplex-Cfd1, Nbp35, and the Cfd1-Nbp35 heterotetrameric complex bound one, three, or four FeS clusters, respectively, upon cluster reconstitution in vitro (13,14). To investigate whether each form of these proteins bound a cluster inside the cell, the supernatants from the primary immunoprecipitation described above were subjected to a second immunoprecipitation with antibody to the other factor, and 55 Fe was measured in the pellets from the primary and secondary immunoprecipitations ( Fig. 1, B and C). Significantly (ϳ4ϫ) more radiolabeled iron co-precipitated with Nbp35 as compared with Cfd1 ( Fig. 1, B and C). Cfd1 and Nbp35 that were specifically precipitated from extracts depleted of the other protein were taken as the noncomplexed or free form of each protein. Yeast co-expressing epitope-tagged Cfd1 (HA) and Nbp35 (Myc) were labeled to steady state with 55 Fe, and cleared extracts were subjected to immunoprecipitation (IP) with protein G-agarose beads alone (Mock) or with epitope-specific antibodies to each protein as indicated. A, left panel, proteins in IP samples were separated by SDS-PAGE and transferred to PVDF membranes, and specific proteins (indicated on the left) were detected by immunoblot (WB). The protein targeted in each primary IP is indicated on the top of the figure. Right panel, proteins remaining in the supernatants after the primary IP (Post-IP Extract) were separated by SDS-PAGE and probed for epitope-tagged Cfd1 or Nbp35 by immunoblot as described above. Relative protein levels were determined by densitometry performed on immunoblot images as described under "Experimental Procedures." Results shown are representative of three independent experiments. B, approximately 80% of the post-IP extracts were subjected to a second round of precipitation with antibodies specific for Nbp35. 55 Fe in precipitates resulting from both the first and the second IPs was quantified by liquid scintillation counting. Nonspecifically associated radioactivity was determined by performing the second precipitation with protein G-agarose beads alone and was subtracted from that obtained with epitope-specific antibodies. Primary, the radioactivity in the first IP. The bars on the right of the graph show the radioactivity in IPs of Nbp35 from extracts that had been immunodepleted with beads alone (Mock) or Cfd1-specific antibodies, as indicated. Inset boxes above each bar show the Nbp35 protein recovered in the second IP. C, analysis as described in B but for Cfd1-associated 55 Fe. The bars on the right of the graph show the radioactivity in IPs of Cfd1 from the indicated immunodepleted extracts. nd, no radioactivity was detected above the background. Inset boxes above each bar show the Cfd1 protein recovered in the second IP. Error bars represent the S.D. of three independent experiments.

Cfd1 Role in Cfd1-Nbp35 Scaffold Complex
Precipitation of Nbp35 from extracts that had been depleted of Cfd1 yielded 55 Fe that was 80% of the level obtained from mockdepleted extracts (Fig. 1B). Thus, the major portion of Nbp35bound 55 Fe (80%) was with free protein, whereas a minor fraction (20%) of the 55 Fe associated with Nbp35 was bound in the Cfd1-Nbp35 heterocomplex at steady state. Precipitation of Cfd1 from Nbp35-depleted extracts yielded no detectable 55 Fe (Fig. 1C), showing that the iron associated with Cfd1 was bound primarily in the Cfd1-Nbp35 heterocomplex. It is important to note that immunoprecipitation of Cfd1 or Nbp35 from mockdepleted extracts yielded 55 Fe at a level comparable with that recovered by primary precipitation (Fig. 1, B and C, compare Primary with Mock). Thus, iron associated with both proteins was stable to the manipulations required for sequential immunoprecipitation and accurately reflected the steady-state distribution of iron among the various forms of these proteins.

55
Fe Labeling Kinetics Reveal Rapid Iron Binding by Nbp35-An FeS scaffold function predicts that Nbp35 will bind and release iron rapidly. As a first step toward addressing these issues, ironstarved yeast were pulsed with 55 Fe at different total added iron concentrations, radioactive iron accumulated over time in cytoplasmic extracts, and Nbp35 was measured. 55 Fe was added at 0.1 M (close to the K m of the high affinity iron transport system consisting of Fet3 and Ftr1 (36)), at 1 M (equivalent to the iron concentration in minimal media (31)), and at 10 M. At 0.1 M iron, maximum 55 Fe accumulation in cells (Fig. 2, upper panel) and near complete medium iron depletion (not shown) were achieved at the earliest time point taken, 15 min. Cells incubated with 1 M 55 Fe accumulated iron up to 60 min after iron addition (Fig. 2, middle panel), the point where medium iron was exhausted (not shown), whereas at 10 M iron, cells continued to accumulate 55 Fe throughout the 90-min experiment and did not significantly deplete iron in the medium (Fig. 2, lower panel).
Binding of 55 Fe to Nbp35 followed closely the pattern of total iron accumulation in yeast cells. At 0.1 M, Nbp35-associated 55 Fe was at maximum by 15 min and remained relatively constant for the remainder of the 90-min experiment (Fig. 2, upper panel). When 55 Fe in the medium was increased to 1 M, Nbp35-associated iron rapidly increased to maximum at 30 min and remain relatively stable out to 90 min (Fig. 2, middle panel). At 10 M medium iron, 55 Fe on Nbp35 increased linearly out to 60 min, at which point it appeared to reach a plateau (Fig. 2, lower panel). The observation that 55 Fe on Nbp35 reached a stable plateau even after total 55 Fe accumulation had stopped suggests either that iron on Nbp35 had reached equilibrium with a stable cellular iron pool or that Nbp35-bound iron was very stable.
Stability of Iron Associated with Cfd1 and Nbp35-A scaffold function also predicts rapid iron turnover on Nbp35 and Cfd1. To address this question, the stability of iron associated with Nbp35 was assessed in a pulse-chase experiment. Iron-starved yeast were pulsed for 30 min at 0.1 M 55 Fe followed by a chase with an excess of nonradioactive iron (see "Experimental Procedures"). Upon the addition of nonradioactive iron to these cells, the 55 Fe immunoprecipitated with Nbp35 decreased rapidly, dropping 40% by 30 min after initiation of the chase (Fig.  3A). Thus, as predicted of an FeS cluster scaffold, the iron associated with Nbp35 was unstable. By contrast, 55 Fe bound to Nar1 and mammalian IRP1 expressed in yeast was very stable during the chase, in each case showing an initial increase in bound 55 Fe (Fig. 3A). Thus, the iron associated with Nbp35 displayed a much more dynamic character than that associated with target proteins, Nar1 and IRP1.
We also examined the stability of the iron bound to Nbp35 and Cfd1 in yeast labeled to steady state with 55 Fe and then transferred to medium containing the membrane-permeant iron chelator BIP. (Cfd1 was labeled poorly in the short 55 Fe pulse of iron-starved yeast but was sufficiently labeled by the long term steady-state procedure to perform this analysis.) BIP inhibits cellular iron uptake in addition to depleting intracellular labile iron pools, rapidly leading to an iron-starved state and inhibition of iron incorporation into protein (33). Iron bound  AUGUST 9, 2013 • VOLUME 288 • NUMBER 32 transiently or in an unstable manner should be rapidly lost from proteins upon exposure of yeast cells to BIP.

Cfd1 Role in Cfd1-Nbp35 Scaffold Complex
The 55 Fe immunoprecipitated with Cfd1 or Nbp35 decreased rapidly upon exposure of yeast to BIP (Fig. 3B). 55 Fe precipitated with Cfd1 or Nbp35 decreased by ϳ80 and 40%, respectively, by 15 min after BIP addition. 55 Fe associated with Cfd1 decreased to background level by 30 min, whereas that associated with Nbp35 dropped to ϳ40% by 30 min after BIP addition. From 30 to 180 min after BIP addition, 55 Fe associated with Nbp35 showed a much slower rate of decrease. The FeS proteins Nar1 and IRP1 showed a more gradual decrease in 55 Fe immunoprecipitated after BIP addition, each showing loss of less than 20% of bound 55 Fe 90 min after BIP addition (Fig. 3B). Taken together, these results support the view of Cfd1 and Nbp35 as FeS cluster scaffolds that are in rapid equilibrium with an intracellular labile iron pool.
Identification of a Heterocomplex-defective cfd1 Mutant-Strains bearing viable cfd1 mutant genes were used to gain further insight into the role of Cfd1 in the heterocomplex with Nbp35. The cfd1-1 allele carries a mutation at the third nucleotide of the translation initiation codon, changing the AUG to an AUA (6). This mutation causes an ϳ90% reduction of Cfd1 level as a result of inefficient translation initiation on the mutant mRNA, yielding a strain that is deficient for normal Cfd1. 6 (Note that N-terminally truncated Cfd1 resulting from translational initiation at in-frame downstream AUG codons is not functional. 6 ) A second mutant allele was selected in a screen of a collection of cfd1 point mutants for altered ability to co-immunoprecipitate with Nbp35. A Cfd1 mutant bearing a change of phenylalanine 199 to serine (Cfd1 F199S ) co-immunoprecipitated with Nbp35 poorly (Fig. 4A) and also co-immunoprecipitated with wild-type Cfd1 less well (Fig. 4B). However, cfd1 F199S yeast were viable, providing a strain with altered heterocomplex stability.
The effect of the cfd1 F199S mutation on indicators of CIA function was investigated and compared with that of the cfd1-1 mutation, which has been reported previously (6). The enzymatic activity of Leu1 and mammalian IRP1 expressed in these yeast and 55 Fe binding to IRP1 and Rli1 were examined. Leu1 activity was inhibited 70% relative to wild-type level in the cfd1-1 strain and 54% in the cfd1 F199S strain (Fig. 5A). IRP1 aconitase activity in these yeast was inhibited by 88% (cfd1-1) and 80% (cfd1 F199S , Fig. 5B). 55 Fe binding to IRP1 reflected enzyme activity, being comparably reduced in each cfd1 mutant strain (Fig. 5C). Finally, 55 Fe binding to Rli1 was strongly inhibited in both mutant strains (Fig. 5D). Interestingly, growth of cfd1-1 and cfd1 F199S strains was not noticeably different from strains bearing wild-type CFD1, indicating that the strongly reduced rate of FeS cluster biogenesis was nonetheless sufficient to support near normal growth. 6 55 Fe binding to the CIA factors, Nbp35 and Nar1, was also examined and compared in the two mutant yeast strains. In yeast labeled to steady state with 55 Fe, binding to Nbp35 was inhibited by 34 and 26% in cfd1-1 and cfd1 F199S mutant strains, whereas 55 Fe binding to Nar1 was inhibited by 44 and 20% in these strains (Fig. 6). Thus, deficiency of Cfd1 activity had a differential inhibitory effect on maturation of FeS proteins in the cytosol, with the stronger effect being on the end stage targets of the CIA pathway (e.g. Rli1). . Aliquots were removed at the indicated times following the addition of nonradioactive iron, and epitope-tagged proteins were immunoprecipitated as described under "Experimental Procedures." 55 Fe that co-precipitated with each protein was determined by liquid scintillation. Data are shown as the percentage of the radioactivity precipitated at the end of the 30-min pulse (time 0). B, cells expressing epitope-tagged Cfd1 (•), Nbp35 (f), Nar1 (), or IRP1 (OE) were labeled to steady state (16 h) with 55 FeCl 3 (1 Ci/ml, ϳ1.2 M total iron concentration). At time 0, BIP (100 M) was added, and aliquots were collected at the indicated times, cell extract was prepared, and each protein was immunoprecipitated and assessed for the presence of 55 Fe by liquid scintillation counting. Radioactivity that nonspecifically bound to protein G-agarose beads was determined and subtracted from each sample. Data are presented as a percentage of radioactivity measured relative to the zero time point. Error bars represent the S.D. of three independent experiments.

FIGURE 4. Effect of Cfd1 F199S mutation on P-loop NTPase interactions.
Yeast expressing HA-tagged Cfd1 or Cfd1 F199S along with either Myc-tagged Nbp35 (A) or Myc-tagged wild-type Cfd1 (B) were grown to mid-log phase and harvested, and cell extracts were prepared and subjected to immunoprecipitation with Myc epitope-specific antibody, as described under "Experimental Procedures." Co-immunoprecipitation of HA-tagged Cfd1 or Cfd1 F199S was detected by immunoblot with anti-HA epitope antibodies. The relative expression of proteins in each strain was determined by Western blot performed on total cell extracts (right panels of A and B).

Cfd1 Mutation Affects Iron Binding and Release from Nbp35-
The results above demonstrate that FeS protein maturation in cfd1-1 and cfd1 F199S strains is defective. To gain insight into the mechanism affected by these mutations, the kinetics of iron binding to Nbp35 and downstream FeS protein targets was examined. To this end, yeast were made iron-deficient and transferred to medium containing 55 FeCl 3 . Aliquots were collected at various times after transfer to 55 Fe-containing medium and analyzed for iron associated with Nbp35, Nar1, or IRP1 (Fig. 7). In cells bearing wild-type Cfd1, 55 Fe rapidly accumulated on Nbp35, reaching a plateau at about 30 min after iron addition (Fig. 7A). The level of radioactive iron associated with Nbp35 remained relatively constant out to 60 min, similar to the results shown in Fig. 2. The Nbp35 labeling kinetics in the cfd1 F199S strain followed a similar pattern to that seen in a strain with wild-type Cfd1 (Fig. 7A). By contrast, the kinetics of iron binding to Nbp35 in the cfd1-1 strain was significantly slower and reached only 50% of the level seen with wild-type Cfd1 at 60 min (Fig. 7A).
Iron binding to Nar1 and IRP1 showed a different pattern as compared with Nbp35. First, 55 Fe binding to Nar1 and IRP1 was linear in yeast bearing wild-type Cfd1 out to at least 1 h (Fig. 7, B  and C). Second, the effect on 55 Fe binding to these proteins by the two Cfd1 mutations was indistinguishable. Although linear, 55 Fe binding to Nar1 was inhibited by ϳ50% (Fig. 7B), whereas 55 Fe binding to IRP1 was inhibited by Ն80% in both cfd1 mutant strains (Fig. 7C). 55 Fe binding to Nbp35 in a yeast strain bearing the Cfd1 F199S mutant was near normal, but downstream targets showed significant defects in iron binding. Therefore it was of interest to determine whether the iron associated with Nbp35 had the same character in the cfd1 F199S strain as in the strain with wildtype Cfd1. To this end, the stability of 55 Fe associated with Nbp35 was examined. Yeast were made iron-starved and then transferred to medium containing 55 Fe for 30 min. At this point, vehicle (none) or BIP at 175 or 250 M was added, and cultures were incubated for 30 min longer.
The amount of 55 Fe that precipitated with Nbp35 from wildtype Cfd1 cells treated with 175 or 250 M BIP was reduced by 26 and 70%, respectively (Fig. 8). Without added BIP, 55 Fe precipitated with Nbp35 increased ϳ2-fold, indicating that the addition of the chelator inhibited iron binding and promoted iron release from Nbp35. By contrast, the 55 Fe precipitated with Nbp35 from both cfd1 mutant strains exposed to 175 and 250 M BIP remained constant relative to the iron associated with the protein at the end of the 30-min labeling period. It should be noted that BIP did block further binding of iron to Nbp35 in the mutant cells (Fig. 8). Thus, the iron bound to Nbp35 in these cfd1 mutant strains was much less labile in the presence of BIP. Given that the cfd1-1 mutation causes severe Cfd1 deficiency, whereas cfd1 F199S encodes a protein that is defective for heterocomplex formation or stability, these results suggest that Cfd1-Nbp35 heterocomplex formation promotes a more labile state of iron bound to Nbp35.
Comparative Analysis of Cfd1 and Nbp35 Mutants-CFD1 and NBP35 are both essential in yeast. However, our data show that yeast can tolerate and survive with severe deficiency in Cfd1 activity. This raises the possibility that Cfd1 provides a catalytic function and that it is more critical that Nbp35 is present in stoichiometric amounts. To explore this possibility, a comparable mutation to that in cfd1-1 changing the AUG translational start codon for synthesis of Nbp35 to AUA was generated, called nbp35 tsm . This mutation caused a similar decrease in Nbp35 expression as seen for Cfd1 from cfd1-1. 6 Like cfd1-1 strains, nbp35 tsm yeast strains were viable, demonstrating that yeast can survive with severely depressed levels of Nbp35 as well. 6 An analysis of indicators of CIA function showed similar defects in FeS cluster biogenesis to that seen for cfd1-1 yeast (Fig. 9). In particular, aconitase activity of IRP1 was depressed ϳ80%, and 55 Fe binding to Nar1 was inhibited ϳ50% in the nbp35 tsm strain. These findings are consistent with the notion that Cfd1 and Nbp35 function in a heterocomplex where depletion of either protein would reduce the level of this heterocomplex and cause equivalent decreases in FeS cluster assembly.
To further investigate the equivalence of these CIA factors, we constructed the F251S mutation in NBP35. Phe-251 is the equivalent residue to Phe-199 in Cfd1 (Fig. 10A). The F251S mutation in Nbp35 was lethal, in contrast to cfd1 F199S . Similar to Cfd1 F199S , Nbp35 F251S was defective for heterocomplex formation/stability and did not co-immunoprecipitate with wildtype Nbp35 (Fig. 11). Moreover, Nbp35 F251S failed to restore FeS cluster assembly in a nbp35 tsm strain as judged from lack of conversion of IRP1 to c-aconitase (Fig. 9). Although these results support the notion that Cfd1 and Nbp35 function in a heterocomplex, they also illustrate the nonequivalence of these factors. Moreover, yeast tolerate deficiency in Cfd1 to a greater degree, suggesting a more fundamental cellular role for Nbp35.

DISCUSSION
All known FeS cluster assembly systems are built around a common set of activities, with the initial assembly of a labile FeS cluster on a scaffold protein a key early step (37). Previous work had shown that when co-expressed in Escherichia coli, Cfd1 and Nbp35 form a heterotetrameric complex that bound four [4Fe-4S] clusters upon reconstitution in vitro (14). Here we report that slightly less than half (40%) of Cfd1 and Nbp35 was FIGURE 6. FeS cluster assembly on Nbp35 and Nar1 in cfd1 mutant yeast. Wild-type and cfd1 mutant strains were grown to mid-log phase in medium containing 55 FeCl 3 and harvested, and whole cell extracts were prepared. Epitope-tagged Nbp35 (A) or Nar1 (B) was immunoprecipitated, and bound 55 Fe was measured by liquid scintillation counting. Error bars represent the S.D. of three independent experiments.

Cfd1 Role in Cfd1-Nbp35 Scaffold Complex
in a complex with each other in logarithmically growing yeast. In yeast grown with 55 Fe, radioactivity was detected with both forms of Nbp35, but only with Cfd1 when bound in the heterocomplex. Significantly, at steady state more 55 Fe was associated with free Nbp35, 80% of total. Moreover, little 55 Fe was detected with Cfd1 in iron-starved yeast given a short pulse of 55 Fe, 6 indicating that nearly all Nbp35-bound 55 Fe in these ironstarved conditions was on free protein. A possible explanation for these results is that Cfd1 binding destabilizes the FeS clusters on Nbp35 to facilitate transfer to target proteins, perhaps making it difficult to capture 55 Fe bound to the Cfd1-Nbp35 complex in iron-starved cells because of a heightened demand for cluster assembly. The rapid decay of 55 Fe from Cfd1 and Nbp35 and the observation that Cfd1 deficiency or mutation (e.g. Cfd1 F199S ) significantly slowed 55 Fe release from Nbp35 support this view. However, Cfd1 deficiency also reduced 55 Fe binding to Nbp35, suggesting that Cfd1 aids in cluster assembly on Nbp35 as well. These observations are consistent with a model in which the Cfd1-Nbp35 heterocomplex is the platform for assembly of labile FeS clusters for transfer to apo proteins. However, at present we cannot rule out that Cfd1 facilitates assembly of both labile and stable FeS clusters on Nbp35.
FeS occupancy on the Cfd1-Nbp35 scaffold was expected to be dynamic and short lived. Indeed 55 Fe on Cfd1 and Nbp35 turned over relatively quickly, with a half-life significantly shorter than that observed for downstream target FeS proteins Nar1 and mammalian IRP1 (Fig. 3). The half-life of 55 Fe on the latter proteins was on the order of hours, whereas essentially all 55 Fe associated with Cfd1 and half of that with Nbp35 was lost in 30 min or less. Interestingly, 55 Fe precipitated with Nar1 and FIGURE 9. Analysis of FeS cluster assembly in the nbp35 tsm mutant strain. A, nbp35 tsm yeast were transformed with IRP1 and grown to mid-log phase, and aconitase activity was measured in cleared cell extracts. To look at Nbp35 F251S functionality, a nbp35 tsm strain was transformed with nbp35 F251S in addition to IRP1, and aconitase was measured in cleared cell extracts (nbp35 tsm ϩ nbp35 F251S ). B, wild-type or nbp35 tsm yeast were transformed with Myc epitope-tagged Nar1. The resulting cells were grown overnight in SD medium containing 55 FeCl 3 . Cells were harvested and washed, and cleared cell extracts were subjected to immunoprecipitation with Myc-specific antibody. Radioactivity that co-precipitated with Nar1 was detected by liquid scintillation. Results are representative of three independent experiments.  (14) is underlined beneath the Af226 sequence. B, ribbon diagram of the C␣ backbone from the crystal structure of the Af226 homodimer (PDB ID: 3KB1). The side-chain aromatic ring of Phe-194 from each monomer is shown as orange sticks, revealing the stacking interaction adopted in the crystal structure. The side chains from the CX 2 C Cys residues are also shown as yellow sticks. Note that these residues coordinate a zinc atom in the crystal (not shown). IRP1 routinely increased by 10 -20% in the initial phases of the chase with nonradioactive iron in pulse-chase experiments (Fig. 3A). Considering that downstream targets such as Nar1 and IRP1 receive preassembled clusters from the Cfd1-Nbp35 scaffold, a pool of Nbp35-bound 55 FeS would exist at the moment of nonradioactive iron addition. These 55 Fe-labeled clusters would continued to be transferred to target proteins until fully exhausted, the 55 Fe being associated with Nbp35 and the Cfd1-Nbp35 complex being replaced by nonradioactive iron from the available pool. Thus, current models for FeS cluster biogenesis involving protein scaffolds predict this result. Importantly, these results support the view that the Cfd1-Nbp35 heterocomplex acquired iron from a dynamic cellular pool that was rapidly diluted by newly transported iron.
Nbp35 has the capacity to bind a ferredoxin-like [4Fe-4S] cluster at the N terminus of each monomer and a single [4Fe-4S] cluster bridging an Nbp35 dimer through the conserved CX 2 C motif near the C terminus of each monomer (13). It has been suggested that the N-terminal ferredoxin-like cluster on Nbp35 is structural and is not subject to transfer to downstream targets, whereas the C-terminal bridged clusters are for transfer to target proteins (13). When yeast were labeled to steady state with 55 Fe followed by exposure to the cell-permeant chelator BIP, ϳ50% of the 55 Fe bound to Nbp35 was rapidly lost, whereas a nearly equal portion of 55 Fe was more stably associated with the protein, persisting on Nbp35 for as long as 3 h in the presence of chelator (Fig. 4A). Although we cannot completely exclude the possibility that both clusters on Nbp35 are labile or even transferred to downstream targets, the kinetic character of Nbp35-bound iron supports the notion of different roles for the FeS clusters on this protein.
Cfd1 also bound a [4Fe-4S] cluster upon reconstitution in vitro, presumably coordinated through its C-terminal CX 2 C motif in a Cfd1 dimer (13). However, in vivo free Cfd1 did not contain detectable FeS, suggesting that it was only competent to associate with clusters when part of a complex with Nbp35. Given the inhibitory effect of Cfd1 deficiency on iron binding to Nbp35, a plausible model then is that Cfd1 interaction with Nbp35 is required for binding of the labile bridging FeS clusters. In support of this view, the C-terminal CX 2 C motif in both proteins is required for function as well as for heterocomplex formation (14). The F199S mutation also destabilized the Cfd1-Nbp35 heterocomplex, but in a short pulse of cfd1 F199S strains, Nbp35 accumulated 55 Fe similar to that seen for wild-type CFD1 yeast (Fig. 7). Because Cfd1 F199S did not co-precipitate with Nbp35, this raises the possibility that only Nbp35 directly binds FeS in the heterocomplex, whereas interaction with Cfd1 changes the character of the cluster-coordinating centers on Nbp35, allowing for more efficient assembly and transfer of FeS clusters to target proteins.
In addition to Cfd1, maximum cluster assembly on Nbp35 requires mitochondrial ISC and ISC export systems and the CIA factors Dre2-Tah18 (2,11). The functions of ISC and Dre2-Tah18 in cytosolic FeS cluster biogenesis have been placed upstream of the Cfd1-Nbp35 interaction in the CIA pathway (11). It is tempting to speculate that mitochondrial ISC, ISC export, and Dre2-Tah18 act together to facilitate assembly of the FeS cluster at the N terminus of Nbp35, whereas Cfd1 is needed for assembly of the C-terminal bridging cluster(s). An intriguing possibility is that assembly of the N-terminal cluster on Nbp35 is a requisite step preceding the assembly of the bridging clusters. However, mutation of either N-terminal or C-terminal cluster ligands impaired the ability of Nbp35 to bind iron in yeast cells, suggesting cooperativity between these sites for cluster assembly and/or stability (14). Further work will be necessary to determine the coordination of FeS cluster assembly at the two sites on Nbp35 and the specific role(s) played by Cfd1 in this process.
A conserved feature among ApbC/Nbp35 protein family members is the presence of a phenylalanine residue (equivalent to Phe-199 in Cfd1) located two residues upstream of the C-terminal CX 2 C cysteine cluster (Fig. 10A). In fact, the crystal structure of Af226, an ApbC/Nbp35 family member of unknown function from Archaeoglobus fulgidus (Protein Data Bank (PDB) ID: 3KB1), shows this conserved Phe residue to be at the homodimer interface, where it forms a bond through stacking with the counterpart residue in the opposing monomer (Fig.  10B). Mutation of this Phe residue to Ser in either Cfd1 or Nbp35 (F199S and F251S, respectively) negatively affected coimmunoprecipitated with self or with the other protein, indicating weakened homodimer and/or heterocomplex. The F251S mutation in Nbp35 was lethal, whereas the comparable mutation in Cfd1 (Cfd1 F199S ) was viable. Thus, the ability of Nbp35 to interact across this interface appears to be more critical than that for Cfd1. A reasonable conclusion from considering these observations is that both Cfd1 and Nbp35 homodimerize in a way consistent with the crystal structure of Af226. In such a scenario, a heterotetrameric complex would form involving other interfaces. Thus, the weak interaction seen for Cfd1 F199S could support bridging cluster assembly on the Nbp35 dimer, but be defective in supporting subsequent transfer of the cluster to downstream targets. Yeast expressing HA-tagged Cfd1 (A) or Nbp35 (B) and either Myc-tagged wildtype Nbp35 or Myc-tagged Nbp35 F251S were grown to mid-log phase and harvested, and cell extracts were prepared and subjected to immunoprecipitation with Myc epitope-specific antibody, as described under "Experimental Procedures." For analysis of Nbp35 F251S interaction with Cfd1, yeast also carried an untagged nbp35 tsm gene to maintain viability. Co-immunoprecipitation of HAtagged Cfd1 or Nbp35 was detected by Western blot with anti-HA epitope antibodies. The relative expression of each protein in each strain was determined by Western blot performed on total cell extracts (right panels of A and B).
Deficiency of either Cfd1 or Nbp35 qualitatively and quantitatively resulted in similar defects in FeS cluster biogenesis, consistent with the notion that the functional form of these proteins is a heterocomplex with each other. The fact that both proteins are essential indicates that each contributes a unique function to this heterocomplex. The results presented here demonstrate that Nbp35 binds the major portion of FeS clusters in the heterocomplex, but that Cfd1 enhances this binding and the release of FeS from the heterocomplex. Interestingly, although Cfd1 is found in metazoans, most fungi, and a number of Archaea, many other organisms (e.g. plants) only carry an Nbp35 homolog for cytosolic FeS cluster biogenesis (19,21,23). Although it is possible that in the latter organisms the functions expressed in Cfd1 were acquired in a single Nbp35 polypeptide, it is intriguing to speculate that Cfd1 evolved from Nbp35 to provide unique functions important in those organisms that carry it. One such function might be to enhance the ability of the CIA system to compete for limiting pools of cellular iron and sulfur, which then allowed this system to become central to cellular iron sensing in metazoans through IRP1 (16). Further insight into this awaits future studies of these P-loop ATPases and the CIA system.